Renal rehabilitation report evaluation method and platform based on big data
Through big data technology, a renal rehabilitation data is collected and processed, an evaluation network library is built, and multi-dimensional evaluation information is generated, which solves the problems of long intervals, poor timeliness and high cost of traditional evaluations, and achieves efficient and low-cost rehabilitation assessment.
Patent Information
- Application Number
- CN202411158771.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-22
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-08-22
AI Technical Summary
Traditional renal rehabilitation assessments have problems such as long intervals, poor timeliness and high cost.
Data on kidney disease patients are collected through the hospital's big data management platform, desensitization and preprocessing are carried out, renal rehabilitation term system is constructed, key features are extracted, evaluation network library is constructed, multi-dimensional evaluation information is generated, and evaluation reports are generated.
Improve evaluation efficiency and timeliness and reduce evaluation costs.
Smart Images

Figure CN118675755B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of health data processing, and in particular to a big data-based kidney rehabilitation report evaluation method and platform. Background Art
[0002] The rehabilitation process of kidney disease is complex and involves data from multiple dimensions, such as clinical indicators, lifestyle, medication use, etc. Traditional kidney rehabilitation assessment mainly relies on regular clinical examinations and doctors' experience and judgment, and has technical problems such as long assessment intervals, poor assessment timeliness, and high assessment costs. Summary of the invention
[0003] The present invention provides a big data-based renal rehabilitation report evaluation method and platform to solve the technical problems of long evaluation intervals, poor evaluation timeliness, and high evaluation costs in the prior art, thereby achieving the technical effect of improving evaluation efficiency and timeliness and reducing evaluation costs.
[0004] In a first aspect, the present invention provides a method for evaluating kidney rehabilitation reports based on big data, wherein the method comprises:
[0005] The kidney disease user data set is collected and acquired through the hospital big data management platform, the kidney disease user data set is desensitized to obtain a kidney disease available data set, and the kidney disease available data set is uploaded to the data processing center; based on the data processing center, the kidney disease available data set is preprocessed and word segmentation text processed to obtain a kidney disease knowledge word segmentation data set, and key features are extracted from the kidney disease knowledge word segmentation data set to obtain a kidney disease key feature data set; a kidney rehabilitation terminology system is established, and a kidney rehabilitation evaluation indicator set is determined according to the kidney rehabilitation terminology system, and the kidney disease key feature data set is annotated, trained and integrated according to the kidney rehabilitation evaluation indicator set to construct a kidney rehabilitation evaluation network library; the target patient's kidney data information is acquired, and the kidney rehabilitation evaluation network library is called and evaluated based on the target patient's kidney data information to obtain kidney rehabilitation multidimensional evaluation information, and text expansion is performed according to the kidney rehabilitation multidimensional evaluation information to generate a target kidney rehabilitation evaluation report.
[0006] In a second aspect, the present invention further provides a renal rehabilitation report evaluation platform based on big data, wherein the platform comprises:
[0007] A data acquisition module, the data acquisition module is used to collect and acquire kidney disease user data sets through the hospital big data management platform, desensitize the kidney disease user data sets, obtain kidney disease available data sets, and upload the kidney disease available data sets to the data processing center; a data feature extraction module, the data feature extraction module is used to perform data preprocessing and word segmentation text processing on the kidney disease available data sets based on the data processing center to obtain a kidney disease knowledge word segmentation data set, extract key features of the kidney disease knowledge word segmentation data set, and obtain a kidney disease key feature data set; an evaluation construction module, the evaluation construction module is used to build a kidney rehabilitation terminology system, determine a kidney rehabilitation evaluation indicator set according to the kidney rehabilitation terminology system, annotate, train and fuse the kidney disease key feature data set according to the kidney rehabilitation evaluation indicator set, and construct a kidney rehabilitation evaluation network library; an evaluation call module, the evaluation call module is used to obtain target patient kidney data information, call and evaluate the kidney rehabilitation evaluation network library based on the target patient kidney data information, obtain kidney rehabilitation multidimensional evaluation information, and perform text expansion based on the kidney rehabilitation multidimensional evaluation information to generate a target kidney rehabilitation evaluation report.
[0008] The present invention discloses a method and platform for evaluating a renal rehabilitation report based on big data, including: collecting a data set of a renal patient through a hospital big data management platform, desensitizing the data set, and uploading the available data set to a data processing center after obtaining the data set. Next, the data processing center performs preprocessing and word segmentation text processing on the available data set of renal disease to generate a knowledge word segmentation data set of renal disease. Then, the word segmentation data set is subjected to key feature extraction to obtain a key feature data set of renal disease. Based on the established renal rehabilitation terminology system, a set of renal rehabilitation evaluation indicators is determined, and these indicators are used to annotate, train and fuse the key feature data set of renal disease to construct a renal rehabilitation evaluation network library. Finally, the renal data information of the target patient is obtained, and the renal rehabilitation evaluation network library is used for evaluation to generate multidimensional evaluation information of renal rehabilitation, and the text is expanded based on this information to finally generate a target renal rehabilitation evaluation report. The method and platform for evaluating a renal rehabilitation report based on big data disclosed by the present invention solve the technical problems of long evaluation intervals, poor evaluation timeliness, and high evaluation costs, and achieve the technical effects of improving evaluation efficiency and timeliness and reducing evaluation costs. BRIEF DESCRIPTION OF THE DRAWINGS
[0009] Figure 1 It is a flow chart of the kidney rehabilitation report evaluation method based on big data of the present invention;
[0010] Figure 2 It is a structural schematic diagram of the kidney rehabilitation report evaluation platform based on big data of the present invention.
[0011] Explanation of the accompanying drawings: data acquisition module 11, data feature extraction module 12, evaluation construction module 13, evaluation calling module 14. DETAILED DESCRIPTION
[0012] The technical solution provided in the embodiments of the present invention is to solve the technical problems of long evaluation interval, poor evaluation timeliness and high evaluation cost in the prior art. The overall idea adopted is as follows:
[0013] First, the kidney disease user data set is collected and obtained through the hospital big data management platform, and the data set is desensitized to obtain the available kidney disease data set, which is then uploaded to the data processing center. Then, based on the data processing center, the available kidney disease data set is preprocessed and word segmentation text processed to obtain a kidney disease knowledge word segmentation data set, and key features are extracted from it to form a kidney disease key feature data set. Then, a kidney rehabilitation terminology system is built, and a set of kidney rehabilitation evaluation indicators is determined based on the system. The kidney disease key feature data set is labeled, trained and fused according to the kidney rehabilitation evaluation indicator set to construct a kidney rehabilitation evaluation network library. Next, the kidney data information of the target patient is obtained, and the kidney rehabilitation evaluation network library is called based on this information for evaluation to obtain multidimensional evaluation information of kidney rehabilitation. Finally, the text is expanded based on the multidimensional evaluation information of kidney rehabilitation to generate a target kidney rehabilitation evaluation report.
[0014] The above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods of the specification to better understand the above technical solution. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments used only to explain the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. In addition, it should be noted that, for the convenience of description, only the parts related to the present invention are shown in the drawings, rather than all of them.
[0015] Embodiment 1
[0016] Figure 1 The figure is a flow chart of a method for evaluating kidney rehabilitation report based on big data of the present invention, wherein the method comprises:
[0017] A kidney disease user data set is collected and acquired through a hospital big data management platform, the kidney disease user data set is desensitized to obtain a kidney disease available data set, and the kidney disease available data set is uploaded to a data processing center.
[0018] Specifically, the relevant data of kidney patients are obtained through the hospital big data management platform, including the patient's medical history, diagnosis information, treatment plan, laboratory test results, imaging data, medication status, etc., to obtain the kidney disease user data set. The kidney disease user data set is the original data containing sensitive information of the patient, so the data is desensitized, including removing or replacing the patient's name, ID number, telephone number and other personal identification information, and blurring the patient's address, date of birth and other sensitive information to ensure that the individual cannot be identified. At the same time, the data is encrypted for storage and transmission to ensure data security.
[0019] Specifically, after confirming that the desensitized data cannot be used to reversely restore personal information and retaining the necessary information required for clinical analysis, the available kidney disease data set is uploaded to the data processing center through a secure data transmission channel.
[0020] In some embodiments, obtaining a kidney disease available dataset comprises:
[0021] Sensitive data of the kidney disease user data set is identified through hospital data application security standards to obtain sensitive data of kidney disease users; application sensitivity evaluation and usage requirement marking are performed on the sensitive data of kidney disease users to determine the data sensitivity level and sensitive data usage requirements; a data desensitization algorithm list is constructed, and based on the data sensitivity level and sensitive data usage requirements, the data desensitization algorithm list is matched in sequence to obtain sensitive data applicability algorithm rules; the kidney disease user data set is desensitized based on the sensitive data applicability algorithm rules to obtain the kidney disease available data set.
[0022] Specifically, first, the hospital data application security standards are used to scan and identify sensitive data in the kidney disease user data set. For example, sensitive data includes the patient's name, ID number, contact information, address, sensitive information in the diagnosis results, etc. Then, the sensitivity level of each sensitive data item is evaluated. For example, the name and ID number have high sensitivity, while the sensitive information in the diagnosis results may need to be partially retained for research and analysis, and personal identity information needs to be completely desensitized.
[0023] Specifically, a list of various desensitization algorithms is constructed, such as partial hiding, encryption, replacement with random numbers, etc., and then the appropriate desensitization algorithm is selected from the algorithm list according to the sensitivity level of the data and the usage requirements, and the corresponding desensitization rules are formulated. For example, the ID card number is partially hidden, encrypted or replaced with a random number for desensitization, while the patient's name is desensitized by replacing it with an anonymous identifier. Through the above-mentioned desensitization algorithm selection, the sensitive information in the data is protected from being leaked without destroying the use value of the data.
[0024] Based on the data processing center, preprocess the available nephropathy dataset and perform word segmentation text processing to obtain a nephropathy knowledge word segmentation dataset, and extract key features from the nephropathy knowledge word segmentation dataset to obtain a nephropathy key feature dataset.
[0025] Specifically, through the data processing center, perform data preprocessing and word segmentation text processing. Among them, data preprocessing includes cleaning, standardizing, transforming, and reshaping data to improve the quality and usability of the data. Among them, word segmentation text processing is a natural language processing method used to decompose continuous text into individual words or phrases, thereby converting text data into a more suitable format for processing.
[0026] In some embodiments, obtaining the nephropathy knowledge word segmentation dataset includes:
[0027] Obtain a data preprocessing program, where the data preprocessing program includes removing non-text content, irrelevant content, stop words, special symbols, correcting outliers, and standardizing text formats; based on the data preprocessing program, perform data standardization processing on the available nephropathy dataset to obtain a nephropathy standard dataset; construct a nephropathy medical dictionary, and use TF-IDF and the nephropathy medical dictionary to perform term word segmentation processing on the nephropathy standard dataset to obtain a nephropathy word segmentation dataset; obtain a noise word list, and based on the noise word list, perform noise word text processing on the nephropathy word segmentation dataset to obtain the nephropathy knowledge word segmentation dataset.
[0028] Specifically, first, perform data preprocessing on the available nephropathy dataset to obtain a nephropathy standard dataset. This data preprocessing is based on the obtained data preprocessing program, which includes removing non-text content, irrelevant content, stop words, special symbols, correcting outliers, and standardizing text formats. Among them, removing non-text content is used to remove non-text data (such as pictures, tables, etc.); removing stop words is used to remove commonly used but meaningless words, such as "de", "shi", etc.; removing irrelevant content is used to remove content irrelevant to nephropathy research (such as other irrelevant test data of patients); removing special symbols is used to remove special symbols and punctuation marks in the text; correcting outliers includes spelling and grammar correction, and normalization correction, which are used to correct spelling mistakes and grammar mistakes in the text and normalize data outside the normal range respectively; standardizing text formats is used to convert all text data into the same encoding format (such as UTF-8) to unify the text format.
[0029] Specifically, the nephropathy medical dictionary is a dictionary containing medical terms and professional vocabulary related to nephropathy. Using the TF-IDF algorithm and the nephropathy medical dictionary, term word segmentation can be performed on the nephropathy standard dataset. Among them, TF-IDF is used to calculate the importance of words to ensure the accurate extraction of key terms.
[0030] Specifically, a noise word list containing noise words such as misspellings and non-kidney disease related words is obtained, and the kidney disease word segmentation dataset is processed according to the noise word list to remove noise words and retain useful information related to kidney disease, so as to further improve data quality and information validity. After the above multi-step processing, a high-quality kidney disease knowledge word segmentation dataset is obtained, which provides high-quality basic data support for subsequent data analysis and model training.
[0031] A renal rehabilitation terminology system is established, and a renal rehabilitation evaluation indicator set is determined based on the renal rehabilitation terminology system. The renal disease key feature data set is labeled, trained and integrated according to the renal rehabilitation evaluation indicator set to construct a renal rehabilitation evaluation network library.
[0032] Optionally, a renal rehabilitation terminology system is built. First, the terminology is classified and the classification criteria of the terminology are determined. Exemplary ones include: renal function indicators, such as glomerular filtration rate (GFR), blood creatinine, urine protein, etc. Description of patient symptoms, such as fatigue, edema, changes in urine volume, etc. Medication, including the name, dosage, mechanism of action, etc. of commonly used drugs. Clinical scores, such as CKD-EPI scores, KDIGO classifications, etc. Rehabilitation status (involving descriptions of stable, worsening, improved conditions, etc.). Then, based on the above classification results, terminology collection and standardization are performed. Exemplary ones include collecting relevant terms from resources such as medical literature, clinical guidelines, and patient records, and standardizing them to ensure terminology consistency and accuracy. Then, each term is defined, and the association between terms is established (such as the association between symptoms and drugs).
[0033] In some embodiments, determining a set of renal rehabilitation assessment indicators comprises:
[0034] Extract related indicators from the renal rehabilitation terminology system, determine a set of renal disease rehabilitation related indicators, perform decentralized processing based on the renal disease rehabilitation related indicator set, and obtain a first set of related indicators; calculate and obtain the covariance matrix of the first set of related indicators, perform eigendecomposition on the covariance matrix, and obtain eigenvalues and corresponding eigenvectors; select principal component information that reaches a preset cumulative contribution rate threshold based on the eigenvalues, and project the renal disease rehabilitation related indicator set to the corresponding eigenvector based on the principal component information to obtain the renal rehabilitation evaluation indicator set after dimensionality reduction.
[0035] Specifically, based on the data dimensionality reduction method, a set of renal rehabilitation evaluation indicators is determined. First, the renal rehabilitation terminology system is extracted based on the significance test method to determine the set of renal disease rehabilitation related indicators; then, the renal disease rehabilitation related indicator set is decentralized, that is, the mean of each indicator is zero. This is a prerequisite for principal component analysis (PCA) to ensure the accuracy of principal component extraction. Among them, the indicators in the renal disease rehabilitation related indicator set include renal function indicators, symptom descriptions, medication information, clinical scores, etc.
[0036] Specifically, the covariance matrix of the decentralized first correlation index set is calculated. The covariance matrix describes the linear relationship between the indicators, and then the covariance matrix is eigendecomposed to obtain the eigenvalues and corresponding eigenvectors. The eigenvalues represent the variance of each principal component, and the eigenvectors describe the linear relationship between the principal component and the original indicator.
[0037] Specifically, according to the preset cumulative contribution rate threshold (such as 80% or 90%), the principal component information that makes the cumulative contribution rate reach or exceed the threshold is selected. The variance of these principal components explains most of the variability in the data. Then, the set of renal rehabilitation-related indicators is projected onto the corresponding eigenvector using the selected principal component information to obtain the set of renal rehabilitation evaluation indicators after dimensionality reduction.
[0038] The reduced dimension index set obtained through the above method steps contains principal component information, which is a linear combination of the original indicators, can avoid multicollinearity problems and improve the accuracy and reliability of the evaluation. Using the reduced dimension kidney rehabilitation evaluation index set to build an evaluation model helps to ensure the performance of the evaluation model while eliminating the interference of invalid or inefficient data.
[0039] In some embodiments, the step of constructing a renal rehabilitation assessment network library comprises:
[0040] The kidney disease key feature data set is associated and classified according to the kidney rehabilitation assessment indicator set to obtain a kidney rehabilitation indicator feature data set; according to the kidney rehabilitation terminology system and term annotation identifier, a kidney rehabilitation term annotation rule is determined, and the kidney rehabilitation indicator feature data set is annotated with terms based on the kidney rehabilitation term annotation rule to obtain a kidney rehabilitation indicator term annotation sample set; the patient's kidney disease characteristic information is obtained, and the patient's kidney disease characteristic information is parameterized to obtain the patient's characteristic calibration parameter information; based on the patient's characteristic calibration parameter information, the kidney rehabilitation indicator term annotation sample set is fused with rehabilitation assessment training to construct the kidney rehabilitation assessment network library.
[0041] Specifically, first, the key feature data set of kidney disease is classified according to the set of kidney rehabilitation evaluation indicators to obtain a kidney rehabilitation indicator feature data set. Exemplarily, association classification is performed through a classification algorithm, such as a decision tree, a support vector machine, etc. Then, the kidney rehabilitation terminology annotation rules are determined according to the kidney rehabilitation terminology system and the term annotation identifier, and the kidney rehabilitation indicator feature data set is annotated with terms to obtain a kidney rehabilitation indicator term annotation sample set. This step may involve some natural language processing technologies, such as named entity recognition, part-of-speech tagging, etc. By simply marking each rehabilitation term, the relationship between the kidney rehabilitation indicator feature data set and the standard terminology is established, which is helpful for subsequent model training.
[0042] Furthermore, the rehabilitation conditions of patients with different characteristics of kidney disease vary greatly, including age, gender, complications, and severe stage of kidney disease, so it is necessary to divide and train their patient characteristics: including obtaining the patient's kidney disease characteristic information and calibrating the parameters of this information to obtain the patient characteristic calibration parameter information. Then, based on the patient characteristic calibration parameter information, the renal rehabilitation indicator terminology annotation sample set is trained to construct a renal rehabilitation evaluation network library.
[0043] Optionally, the renal rehabilitation assessment network library is constructed based on machine learning or deep learning algorithms, such as neural networks, convolutional neural networks, recurrent neural networks, etc. And the renal rehabilitation assessment network library includes calibration parameter information corresponding to different patient characteristics, with a variety of network structures, complexity and network parameters. In other words, based on multiple groups of patient characteristic calibration parameter information corresponding to multiple patients, respective renal rehabilitation assessment networks are constructed and stored as renal rehabilitation assessment network libraries, which are convenient for rapid call in actual use.
[0044] Specifically, constructing different renal rehabilitation assessment networks for different types of patients through the above-mentioned method steps can help improve the efficiency of network training, while reducing the scale and computing power consumption of a single assessment network, which in turn helps to promote it in practical applications.
[0045] In some implementations, constructing the renal rehabilitation assessment network library includes:
[0046] Based on the patient characteristic calibration parameter information, the renal rehabilitation indicator terminology annotation sample set is classified to obtain multiple calibration parameter renal rehabilitation annotation sample sets; the multiple calibration parameter renal rehabilitation annotation sample sets are respectively subjected to rehabilitation assessment training according to the kidney disease key feature data set to generate multiple calibration parameter indicator rehabilitation assessment network sets; the multiple calibration parameter indicator rehabilitation assessment network sets are respectively subjected to average weight network fusion according to the patient characteristic calibration parameter information to obtain multiple calibration parameter renal rehabilitation assessment networks; the multiple calibration parameter renal rehabilitation assessment networks are calibrated and integrated based on the patient characteristic calibration parameter information to construct the renal rehabilitation assessment network library.
[0047] Specifically, first, the patient characteristic calibration parameter information is used as the classification basis, and the annotation labels of the renal rehabilitation indicator terminology annotation sample set are classified and divided to obtain multiple calibration parameter renal rehabilitation annotation sample sets, wherein different calibration parameter renal rehabilitation annotation sample sets correspond to different types of patients. Then, for each calibration parameter renal rehabilitation annotation sample set, rehabilitation assessment training is performed according to the key feature data set of kidney disease to generate multiple calibration parameter indicator rehabilitation assessment network sets. Among them, each set includes an evaluation network of a specific characteristic parameter, exemplarily including an evaluation network for a specific age group, a specific underlying disease, etc.
[0048] Specifically, multiple calibration parameter index rehabilitation evaluation network sets are averagely weighted and fused according to the patient characteristic calibration parameter information. During the fusion process, the index rehabilitation evaluation network of each characteristic parameter is fused with the same index weight. That is, the same weight is given to the evaluation network of the same index to ensure that the influence of each network on the evaluation result in the fusion is balanced. The obtained multiple calibration parameter renal rehabilitation evaluation networks fuse the evaluation information of different characteristic parameters and provide comprehensive rehabilitation evaluation capabilities.
[0049] Furthermore, based on the patient characteristic calibration parameter information, multiple calibration parameter renal rehabilitation assessment networks are calibrated and integrated to further optimize the accuracy and reliability of the assessment network and ensure its adaptability to different patient characteristics in practical applications. The renal rehabilitation assessment network library constructed through the above steps can provide personalized assessment services, support clinicians to develop more scientific rehabilitation plans, and thus improve the rehabilitation effect of patients.
[0050] Obtain target patient kidney data information, call and evaluate the kidney rehabilitation assessment network library based on the target patient kidney data information, obtain kidney rehabilitation multidimensional assessment information, and perform text expansion based on the kidney rehabilitation multidimensional assessment information to generate a target kidney rehabilitation assessment report.
[0051] Specifically, based on the patient characteristic parameters of the target patient, the corresponding renal rehabilitation assessment network is matched in the renal rehabilitation assessment network library to perform multi-dimensional rehabilitation assessment. Among them, the matched renal rehabilitation assessment network is a personalized renal rehabilitation assessment network, which has the highest degree of adaptability to the actual conditions of the target patient, thereby ensuring the accuracy of the obtained assessment information.
[0052] Specifically, text expansion is performed based on the multidimensional assessment information of renal rehabilitation, including calling a preset text expansion template to convert discrete indicative multidimensional assessment information of renal rehabilitation into a structured text with high readability and guiding effect. Exemplarily, the preset text expansion template includes rehabilitation progress assessment, complication risk warning, treatment recommendations (such as drug adjustment, lifestyle changes, etc.) and future rehabilitation plans.
[0053] Furthermore, the method comprises:
[0054] The target renal rehabilitation assessment report is audited and verified to obtain the rehabilitation report audit feedback parameters; based on the rehabilitation report audit feedback parameters, the problem classification and location are performed to determine the model feature parameters to be optimized; based on the model feature parameters to be optimized, the model parameter optimization strategy is determined, and based on the model parameter optimization strategy, the target call rehabilitation assessment network is optimized and updated.
[0055] Specifically, the generated renal rehabilitation assessment report is audited and verified through manual review by professionals or automated review tools. The result of the audit verification is the rehabilitation report audit feedback parameter, which can help you understand the quality of the report and possible problems. Exemplarily, the audit feedback parameters may include result accuracy parameters, model efficiency parameters, etc.
[0056] Specifically, the model feature parameters to be optimized refer to the model parameters that need to be optimized, which are concrete representations of possible problems or deficiencies in the model, including data bias, model error, text generation logic, etc.
[0057] Specifically, the model parameter optimization strategy is determined according to the feature parameters to be optimized. The model parameter optimization strategy includes: adjusting model parameters, improving algorithm structure, introducing new features, etc.
[0058] In summary, the big data-based renal rehabilitation report evaluation method provided by the present invention has the following technical effects:
[0059] The data set of kidney disease patients is collected through the hospital big data management platform, and it is desensitized. After obtaining the available data set, it is uploaded to the data processing center. Next, the data processing center preprocesses the available data set of kidney disease and performs word segmentation text processing to generate a knowledge word segmentation data set of kidney disease. Then, the key features of the word segmentation data set are extracted to obtain the key feature data set of kidney disease. Based on the established kidney rehabilitation terminology system, the set of kidney rehabilitation evaluation indicators is determined, and these indicators are used to annotate, train and fuse the key feature data set of kidney disease to construct a kidney rehabilitation evaluation network library. Finally, the kidney data information of the target patient is obtained, and the kidney rehabilitation evaluation network library is used for evaluation to generate multidimensional evaluation information of kidney rehabilitation, and the text is expanded based on this information to finally generate the target kidney rehabilitation evaluation report. Thereby achieving the technical effect of improving the efficiency and timeliness of evaluation and reducing the cost of evaluation.
[0060] Embodiment 2
[0061] Figure 2 Schematic diagram of the structure of the renal rehabilitation report evaluation platform based on big data of the present invention. For example, Figure 1 The flowchart of the big data-based renal rehabilitation report evaluation method of the present invention can be shown as follows: Figure 2 The structure shown is implemented.
[0062] Based on the same concept as the big data-based renal rehabilitation report evaluation method in the above embodiment, the present invention also provides a big data-based renal rehabilitation report evaluation platform including:
[0063] The data acquisition module 11 is used to collect and acquire a kidney disease user data set through a hospital big data management platform, desensitize the kidney disease user data set, obtain a kidney disease available data set, and upload the kidney disease available data set to a data processing center.
[0064] The data feature extraction module 12 is used to perform data preprocessing and word segmentation text processing on the kidney disease available data set based on the data processing center to obtain a kidney disease knowledge word segmentation data set, and perform key feature extraction on the kidney disease knowledge word segmentation data set to obtain a kidney disease key feature data set.
[0065] The evaluation construction module 13 is used to build a renal rehabilitation terminology system, determine a set of renal rehabilitation evaluation indicators based on the renal rehabilitation terminology system, annotate, train and fuse the renal disease key feature data set according to the renal rehabilitation evaluation indicator set, and build a renal rehabilitation evaluation network library.
[0066] The evaluation calling module 14 is used to obtain the kidney data information of the target patient, call and evaluate the kidney rehabilitation evaluation network library based on the kidney data information of the target patient, obtain multidimensional evaluation information of kidney rehabilitation, and perform text expansion based on the multidimensional evaluation information of kidney rehabilitation to generate a target kidney rehabilitation evaluation report.
[0067] Wherein, the data acquisition module 11 includes:
[0068] The sensitive data identification and acquisition unit is used to perform sensitive data identification on the kidney disease user data set through the hospital data application security standard to obtain the sensitive data of the kidney disease user.
[0069] The sensitive data evaluation and marking unit is used to evaluate the application sensitivity and mark the usage requirements of the sensitive data of the kidney disease user, and determine the data sensitivity level and sensitive data usage requirements.
[0070] The data desensitization algorithm matching unit is used to construct a data desensitization algorithm list, and based on the data sensitivity level and sensitive data usage requirements, match the data desensitization algorithm list in sequence to obtain sensitive data applicability algorithm rules.
[0071] A data desensitization processing unit is used to perform desensitization processing on the kidney disease user data set based on the sensitive data applicability algorithm rules to obtain the kidney disease available data set.
[0072] In some implementations, the data feature extraction module 12 includes:
[0073] The data preprocessing program acquisition unit is used to acquire a data preprocessing program, wherein the data preprocessing program includes non-text content removal, irrelevant content, stop words, special symbol removal, outlier correction and text format standardization.
[0074] A data standardization processing unit is used to perform data standardization processing on the available kidney disease data set based on the data preprocessing program to obtain a kidney disease standard data set.
[0075] The medical dictionary construction unit is used to construct a kidney disease medical dictionary, and use TF-IDF and the kidney disease medical dictionary to perform term segmentation processing on the kidney disease standard data set to obtain a kidney disease segmentation data set.
[0076] The noise word list acquisition unit is used to acquire a noise word list, and perform noise word text processing on the kidney disease word segmentation data set based on the noise word list to obtain the kidney disease knowledge word segmentation data set.
[0077] In some implementations, the evaluation building block 13 includes:
[0078] The renal rehabilitation related indicator extraction unit is used to extract related indicators from the renal rehabilitation terminology system, determine a set of renal disease rehabilitation related indicators, perform decentralized processing based on the set of renal disease rehabilitation related indicators, and obtain a first set of related indicators.
[0079] The covariance matrix calculation unit is used to calculate and obtain the covariance matrix of the first correlation indicator set, perform eigendecomposition on the covariance matrix, and obtain eigenvalues and corresponding eigenvectors.
[0080] The principal component selection and projection unit is used to select the principal component information that reaches the preset cumulative contribution rate threshold according to the eigenvalue, and project the kidney disease rehabilitation related indicator set to the corresponding eigenvector based on the principal component information to obtain the kidney rehabilitation evaluation indicator set after dimensionality reduction.
[0081] In some implementations, the evaluation building block 13 includes:
[0082] The renal rehabilitation index characteristic data set construction unit is used to associate and classify the renal disease key characteristic data set according to the renal rehabilitation evaluation index set to obtain the renal rehabilitation index characteristic data set.
[0083] A renal rehabilitation terminology labeling rule determination and execution unit is used to determine the renal rehabilitation terminology labeling rules according to the renal rehabilitation terminology system and terminology labeling identifier, and to label the renal rehabilitation indicator feature data set based on the renal rehabilitation terminology labeling rules to obtain a renal rehabilitation indicator terminology labeling sample set.
[0084] The patient characteristic information parameter calibration unit is used to obtain the patient's renal disease characteristic information, perform parameter calibration on the patient's renal disease characteristic information, and obtain the patient characteristic calibration parameter information.
[0085] A rehabilitation assessment training fusion unit is used to perform rehabilitation assessment training fusion on the renal rehabilitation indicator term labeling sample set based on the patient characteristic calibration parameter information to construct the renal rehabilitation assessment network library.
[0086] In some implementations, the rehabilitation assessment training fusion unit further includes:
[0087] The calibration parameter renal rehabilitation annotated sample set classification unit is used to classify the renal rehabilitation indicator term annotated sample set based on the patient characteristic calibration parameter information to obtain multiple calibration parameter renal rehabilitation annotated sample sets.
[0088] The rehabilitation assessment training unit is used to perform rehabilitation assessment training on the multiple calibration parameter kidney rehabilitation annotation sample sets according to the kidney disease key feature data set to generate multiple calibration parameter indicator rehabilitation assessment network sets.
[0089] The weighted network fusion unit is used to perform weighted network fusion on the multiple calibration parameter index rehabilitation assessment network sets according to the patient characteristic calibration parameter information to obtain multiple calibration parameter kidney rehabilitation assessment networks.
[0090] The renal rehabilitation assessment network calibration integration unit is used to calibrate and integrate the multiple calibration parameter renal rehabilitation assessment networks based on the patient characteristic calibration parameter information to construct the renal rehabilitation assessment network library.
[0091] In some implementations, the platform further includes:
[0092] The rehabilitation report review and verification unit is used to review and verify the target kidney rehabilitation assessment report to obtain rehabilitation report review feedback parameters.
[0093] The problem classification and positioning unit is used to classify and locate problems based on the rehabilitation report review feedback parameters, and determine the characteristic parameters of the model to be optimized.
[0094] The model parameter optimization strategy determination unit is used to determine the model parameter optimization strategy according to the characteristic parameters to be optimized of the model.
[0095] The rehabilitation assessment network optimization and updating unit is used to optimize and update the target-invoked rehabilitation assessment network based on the model parameter optimization strategy.
[0096] It should be understood that the embodiments mentioned in this specification focus on their differences from other embodiments. The specific embodiments in the aforementioned embodiment one are also applicable to the big data-based renal rehabilitation report evaluation platform described in embodiment two. For the sake of brevity of the specification, they will not be further elaborated here.
[0097] It should be understood that the embodiments disclosed in the present invention and the above description can enable those skilled in the art to use the present invention to implement the present invention. At the same time, the present invention is not limited to the above-mentioned embodiments. It should be understood that those skilled in the art can still modify the technical solutions recorded in the above-mentioned embodiments, or replace some of the technical features therein by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.
Claims
1. A big data-based renal rehabilitation report evaluation method, characterized in that: The method comprises: Acquire a kidney disease user data set through a hospital big data management platform, desensitize the kidney disease user data set to obtain a kidney disease available data set, and upload the kidney disease available data set to a data processing center; Based on the data processing center, the available kidney disease data set is subjected to data preprocessing and word segmentation text processing to obtain a kidney disease knowledge word segmentation data set, and key features are extracted from the kidney disease knowledge word segmentation data set to obtain a kidney disease key feature data set; Build a renal rehabilitation terminology system, determine a renal rehabilitation evaluation index set based on the renal rehabilitation terminology system, annotate, train and fuse the renal disease key feature data set according to the renal rehabilitation evaluation index set, and build a renal rehabilitation evaluation network library; Acquire the target patient's kidney data information, call and evaluate the kidney rehabilitation assessment network library based on the target patient's kidney data information to obtain kidney rehabilitation multidimensional assessment information, and perform text expansion based on the kidney rehabilitation multidimensional assessment information to generate a target kidney rehabilitation assessment report; The determination of a set of renal rehabilitation assessment indicators includes: Extracting related indicators from the renal rehabilitation terminology system to determine a set of related indicators for renal disease rehabilitation, performing decentralization processing based on the set of related indicators for renal disease rehabilitation to obtain a first set of related indicators; Calculate and obtain the covariance matrix of the first association indicator set, perform eigendecomposition on the covariance matrix, and obtain eigenvalues and corresponding eigenvectors; Selecting principal component information that reaches a preset cumulative contribution rate threshold according to the eigenvalues, projecting the kidney disease rehabilitation-related indicator set to the corresponding eigenvector based on the principal component information, and obtaining the kidney rehabilitation evaluation indicator set after dimensionality reduction; The construction of the renal rehabilitation assessment network library includes: Associatively classifying the kidney disease key feature data set according to the kidney rehabilitation assessment index set to obtain a kidney rehabilitation index feature data set; Determine a renal rehabilitation terminology annotation rule according to the renal rehabilitation terminology system and terminology annotation identifier, and perform terminology annotation on the renal rehabilitation indicator feature data set based on the renal rehabilitation terminology annotation rule to obtain a renal rehabilitation indicator terminology annotation sample set; Acquiring patient renal disease characteristic information, performing parameter calibration on the patient renal disease characteristic information, and obtaining patient characteristic calibration parameter information; Based on the patient characteristic calibration parameter information, the renal rehabilitation indicator terminology annotation sample set is subjected to rehabilitation assessment training fusion to construct the renal rehabilitation assessment network library; The step of constructing the renal rehabilitation assessment network library comprises: Classifying the renal rehabilitation indicator terminology labeled sample set based on the patient characteristic calibration parameter information to obtain a plurality of calibration parameter renal rehabilitation labeled sample sets; Perform rehabilitation assessment training on the multiple calibration parameter kidney rehabilitation annotated sample sets according to the kidney disease key feature data set to generate multiple calibration parameter indicator rehabilitation assessment network sets; Performing weighted network fusion on the multiple calibration parameter index rehabilitation assessment network sets respectively according to the patient characteristic calibration parameter information to obtain multiple calibration parameter kidney rehabilitation assessment networks; The plurality of calibration parameter renal rehabilitation assessment networks are calibrated and integrated based on the patient characteristic calibration parameter information to construct the renal rehabilitation assessment network library.
2. The big data-based renal rehabilitation report evaluation method according to claim 1, characterized in that: The available kidney disease data sets obtained include: Performing sensitive data identification on the kidney disease user data set by using hospital data application security standards to obtain sensitive data of the kidney disease user; Conduct application sensitivity assessment and usage requirement marking on the sensitive data of the kidney disease user to determine the data sensitivity level and usage requirements of the sensitive data; Construct a data desensitization algorithm list, and match the data desensitization algorithm list in turn based on the data sensitivity level and sensitive data usage requirements to obtain algorithm rules for sensitive data applicability; The kidney disease user data set is desensitized based on the sensitive data applicability algorithm rules to obtain the kidney disease available data set.
3. The big data-based renal rehabilitation report evaluation method according to claim 1, characterized in that: The obtained kidney disease knowledge word segmentation dataset includes: Obtaining a data preprocessing procedure, wherein the data preprocessing procedure includes removing non-text content, irrelevant content, stop words, special symbols, correcting outliers, and standardizing text formats; Performing data standardization processing on the available kidney disease data set based on the data preprocessing program to obtain a kidney disease standard data set; Constructing a kidney disease medical dictionary, and performing term segmentation processing on the kidney disease standard data set using TF-IDF and the kidney disease medical dictionary to obtain a kidney disease segmentation data set; A noise word list is obtained, and noise word text processing is performed on the kidney disease word segmentation dataset based on the noise word list to obtain the kidney disease knowledge word segmentation dataset.
4. The big data-based renal rehabilitation report evaluation method according to claim 1, characterized in that: The method comprises: Review and verify the target kidney rehabilitation assessment report to obtain the rehabilitation report review feedback parameters; Based on the rehabilitation report review feedback parameters, problem classification and location are performed to determine the characteristic parameters of the model to be optimized; According to the characteristic parameters to be optimized of the model, a model parameter optimization strategy is determined, and based on the model parameter optimization strategy, the target calling renal rehabilitation assessment network library is optimized and updated.
5. A big data-based renal rehabilitation report evaluation platform, characterized by: The platform includes: A data acquisition module, the data acquisition module is used to collect and acquire a kidney disease user data set through a hospital big data management platform, perform desensitization processing on the kidney disease user data set, obtain a kidney disease available data set, and upload the kidney disease available data set to a data processing center; A data feature extraction module, wherein the data feature extraction module is used to perform data preprocessing and word segmentation text processing on the available kidney disease data set based on the data processing center to obtain a kidney disease knowledge word segmentation data set, and to extract key features from the kidney disease knowledge word segmentation data set to obtain a kidney disease key feature data set; An evaluation construction module, wherein the evaluation construction module is used to build a renal rehabilitation terminology system, determine a renal rehabilitation evaluation indicator set according to the renal rehabilitation terminology system, annotate, train and fuse the renal disease key feature data set according to the renal rehabilitation evaluation indicator set, and build a renal rehabilitation evaluation network library; An evaluation calling module, the evaluation calling module is used to obtain kidney data information of a target patient, call and evaluate the kidney rehabilitation evaluation network library based on the kidney data information of the target patient, obtain multi-dimensional evaluation information of kidney rehabilitation, and perform text expansion according to the multi-dimensional evaluation information of kidney rehabilitation to generate a target kidney rehabilitation evaluation report; The evaluation building block also includes: A renal rehabilitation related indicator extraction unit, configured to extract related indicators from the renal rehabilitation terminology system, determine a set of renal disease rehabilitation related indicators, and perform decentralized processing based on the set of renal disease rehabilitation related indicators to obtain a first set of related indicators; A covariance matrix calculation unit, used for calculating and obtaining the covariance matrix of the first correlation indicator set, performing eigendecomposition on the covariance matrix, and obtaining eigenvalues and corresponding eigenvectors; A principal component selection and projection unit, configured to select principal component information reaching a preset cumulative contribution rate threshold according to the eigenvalue, and project the set of kidney disease rehabilitation-related indicators to the corresponding eigenvector based on the principal component information to obtain the set of kidney rehabilitation evaluation indicators after dimensionality reduction; The evaluation building block also includes: A renal rehabilitation index characteristic data set construction unit, used for associating and classifying the renal disease key characteristic data set according to the renal rehabilitation evaluation index set to obtain a renal rehabilitation index characteristic data set; A renal rehabilitation terminology annotation rule determination and execution unit, used to determine the renal rehabilitation terminology annotation rule according to the renal rehabilitation terminology system and the terminology annotation identifier, and perform terminology annotation on the renal rehabilitation indicator feature data set based on the renal rehabilitation terminology annotation rule to obtain a renal rehabilitation indicator terminology annotation sample set; A patient characteristic information parameter calibration unit, used to obtain the patient's renal disease characteristic information, perform parameter calibration on the patient's renal disease characteristic information, and obtain patient characteristic calibration parameter information; A rehabilitation assessment training fusion unit, used to perform rehabilitation assessment training fusion on the renal rehabilitation indicator terminology annotation sample set based on the patient characteristic calibration parameter information, and construct the renal rehabilitation assessment network library; The rehabilitation assessment training fusion unit also includes: A calibration parameter renal rehabilitation annotated sample set classification unit is used to classify the renal rehabilitation indicator term annotated sample set based on the patient characteristic calibration parameter information to obtain multiple calibration parameter renal rehabilitation annotated sample sets; A rehabilitation assessment training unit, used to perform rehabilitation assessment training on the multiple calibration parameter kidney rehabilitation annotation sample sets according to the kidney disease key feature data set, and generate multiple calibration parameter indicator rehabilitation assessment network sets; A weighted network fusion unit, used to perform weighted network fusion on the plurality of calibration parameter index rehabilitation assessment network sets respectively according to the patient characteristic calibration parameter information, to obtain a plurality of calibration parameter kidney rehabilitation assessment networks; The renal rehabilitation assessment network calibration integration unit is used to calibrate and integrate the multiple calibration parameter renal rehabilitation assessment networks based on the patient characteristic calibration parameter information to construct the renal rehabilitation assessment network library.
